Grad school or a job. A startup or a steady salary. Pathwise models the tradeoffs across time instead of handing you a pros-and-cons list — and it always shows its uncertainty instead of faking confidence.
Optional. Without a key, Pathwise uses its built-in template narrator — the tradeoff numbers themselves are identical either way, since those come from the rules engine, not the LLM.
Deterministic and auditable. Same inputs always produce the same scores. This is where the actual "reasoning structure" lives — four lenses × three time horizons, with confidence bands instead of fake precision. No hallucination risk because nothing here is generated.
The rules engine can't read "my parents really want me to play it safe" in your free-text box and adjust its tone. An LLM can. It only ever explains the numbers already computed — it's instructed never to override them, and the app works fully without it.
Pathwise will never output a field called "recommendation." It computes a weighted lean, always shows the runner-up's strongest dimension, and the actual choice — the one with real consequences — stays entirely with you.